MBON Acoustic Indices Study

Research Question

Can acoustic indices predict biological community metrics in estuarine environments?

Location: 3 stations, May River, South Carolina
Period: 2021 (full year)
Responses: 9 community metrics
- Fish: activity, richness, presence
- Dolphins: echolocation, burst pulse, whistle, total activity, presence
- Vessels: presence
Predictors: ~60 acoustic indices (candidates)

Data Overview

13,102 observations (2-hour temporal bins)
4 data sources aligned to common resolution:

Source	Description
Detections	Manual annotations of fish/dolphin/vessel presence
Environment	Temperature, depth (sensor data)
Acoustic Indices	~60 indices across 5 categories
SPL	Sound pressure levels

Index categories: Amplitude, Complexity, Diversity, Spectral, Temporal
Temporal structure: station / month / day / hour

Pipeline Overview

Stage 00: Data Alignment (4 sources → 2-hour bins)
    ↓
Stage 01: Index Reduction (60 → 14 via correlation/VIF)
    ↓
Stage 02-03: Response Variables + Feature Engineering
    ↓
Stage 05: GAMM modeling (mgcv::bam)

Model choice: GAMM (Generalized Additive Mixed Model)

Allows non-linear (smooth) relationships between predictors and response
Increasingly common in ecological literature for this type of study
Preliminary comparison with GLMM showed strong GAMM preference (ΔAIC > 6000)

Model Specifications

GAMM (mgcv::bam)

Smooth terms (k=5) for indices & covariates
Cyclic splines for hour, day-of-year
Random effects: station, month
AR1 via rho parameter

Model types:

Negative binomial (nbinom2) for count responses
Binomial for presence/absence responses

Question 1: Index Reduction

Is our approach appropriate? Should we reduce further?

Index Reduction: What We Did

Step 1: Correlation pruning

Removed one index from each pair with |r| > 0.6
Result: 60 → 17 indices

Step 2: VIF screening

Iteratively removed indices with VIF > 2
Result: 17 → 14 indices

Outcome:

14 indices retained
All 5 categories preserved (Amplitude, Complexity, Diversity, Spectral, Temporal)

Index Reduction: Concerns

Is 14 indices too many?
Is correlation + VIF the right approach?
- Alternatives: PCA, LASSO, elastic net (others?)
Model shrinkage removed 4 more
- GAMM select=TRUE shrunk ADI, BioEnergy, EPS_KURT, MEANt to ~zero
- Should we report 14 predictors or 10 “effective” predictors?
- Should we formalize this as a two-stage approach? Remove the “shrunk” indices and rerun?

Index Reduction: Questions for Discussion

Q1: Is correlation + VIF standard practice, or is there a more appropriate approach?

Q2: Given that GAMM shrinkage removed 4 indices, should we adopt a two-stage approach (VIF → model-based selection)?

Q3: 14 predictors for 13K observations — but effective sample size is lower due to temporal autocorrelation.

Question 2: Modeling Results

What are our results telling us? Any concerns?

GAMM Results: Significant Predictors

Term	EDF	p-value	Interpretation
hour_of_day	8.24	<0.001	Strong diel pattern
ACI	2.66	<0.001	Non-linear, positive
BI	2.82	<0.001	Non-linear, negative
EAS	3.08	<0.001	Non-linear
VARt	2.94	<0.001	Non-linear
depth	1.00	<0.001	Linear, negative

Shrunk away (not significant): ADI, BioEnergy, EPS_KURT, MEANt

GAMM Smooth Plots: Overview

Smooth Zoom: hour_of_day

Observations:

Strong diel pattern (EDF = 8.2)
Peak activity ~8 PM (hour 20)
Lowest ~10 AM (hour 10)

Validation:

Matches known fish calling behavior
Model is capturing real biology

Smooth Zoom: BI (negative relationship)

Observation:

Higher BI → less fish activity
Counterintuitive?

Possible explanations:

BI elevated when other sources dominate (snapping shrimp?)
Fish call when BI is lower
Seasonal confounding?

Question: Ecologically interpretable or artifact?

Smooth Zoom: VARt (non-linear)

Observation:

“Goldilocks” relationship
Fish activity peaks at intermediate VARt
Drops at both extremes

Implication:

This non-linear pattern is common in ecology
GAMM smooths capture it naturally

Unexpected Results

Temperature:

NOT significant in GAMM (p = 0.12)

Day of year:

NOT significant in GAMM (p = 0.18)
Despite visible seasonality in data →

Hypothesis: Acoustic indices absorb the seasonal/temperature signal?

Methodological Concerns

10 of 14 indices significant
- Genuine signal or overfitting?
- Large sample size (13K) means small effects can be significant
AR1 autocorrelation
- Currently using fixed rho = 0.6
- Should we estimate rho from the data?
Indices absorbing environmental signal?
- Temperature and seasonality not significant
- But indices vary with both — collinearity concern?

Question 3: Validation Approach

How much validation do we need for a journal paper?

Inference vs Prediction: Options

Approach	What it means	Pros	Cons
Inference only	Full-data GAMM, report coefficients	Simpler, answers “are there relationships?”	Reviewers may question generalizability
Inference + light CV	Add leave-one-station-out validation	Shows relationships are robust	Slightly more work
Full predictive	Extensive CV, prediction metrics	Strongest for “applied monitoring” framing	Scope creep

Our tentative plan: Inference + light CV

Primary: Understanding which indices relate to community metrics
Supporting: Simple CV showing relationships generalize
Future work: Operational prediction applications

Summary: Questions for Discussion

Index reduction: Is correlation + VIF appropriate? Should we reduce further, or is model shrinkage sufficient?
Interpretation: Temperature/seasonality not significant — absorbed by indices, or a problem?
Validation: Is inference + light CV sufficient for publication? What kind of CV would reviewers expect?
Next steps: Expand to all 9 responses? Additional diagnostics?

Additional Context

Pilot mode: Results shown are for fish_activity only — will expand to all 9 responses
MEANt: No real variation in raw data (numerical noise ~10⁻¹⁹) — model correctly shrunk it away
Repository: [link] — full specs, code, and data pipeline available